Cancers of the head and neck are commonly unresponsive to chemotherapy and radiation therapy. Some are inoperable due to their location or nature of infiltration. Others are surgically tractable but require mutilating surgery that results in impaired sensory and communication capability. The physical accessibility of these tumors and their resistance to standard non-surgical treatments make them ideal candidates for new therapeutic approaches such as gene therapy. The transduction of cells with a herpes virus thymidine kinase gene (HSVtk) sensitizes the transduced cells to ganciclovir. Also, nearby cells that lack the HSVtk, are killed b ganciclovir administration. This phenomenon is known as a 'bystander effect." This application builds upon our observations which show that metabolic cooperation (i.e. the transfer of small cytotoxic molecules between adjacent cells via gap junctions) plays a significant role in the killing of nearby cells that lack HSVtk. In this application, we propose to enhance the bystander effect to make HSVtk gene therapy more effective for head and neck cancer. The premise is that if the HSVtk enzyme activity is elevated, the metabolism of ganciclovir to its phosphorylated cytotoxic derivative will be increased, as will its intracellular concentration, leading to a more extensive bystander effect. This issue will be addressed in several ways: I) We will deliver multiple copies of the HSVtk gene to cells in vitro and tumors in vivo, rather than the single integrated copy resulting from HSVtk retroviral infection. The delivery system of choice will be a liposome/DNA complex. The efficiency of delivery and the relationship between gene copy number, level of expression and extent of the bystander effect will be tested in vitro and in vivo. 2) We will test viral promoters and keratin K6 and K16 promoters to identify those that will transcribe most efficiently in the target cells. Again, they will be tested with respect to extent of bystander effect in vitro and in vivo. 3) We will use a genetic approach with random PCR mutagenesis to produce and select for mutant HSVtk genes and proteins that metabolize ganciclovir more efficiently than wild type. Mutant enzymes will be characterized, and their impact on the bystander effect will be assessed.